Stop Press
(although, like, that can only be done before something gets published, I guess ;-)There's a chance this is down to an IP address change on Adobe's end, and a DNS caching issue on my end. Stay tuned...
...Stop the press, take to it with a sledgehammer, throw press away
Well I owe Adobe an apology! It was indeed mostly an IP address chance at their end, and DNS caching on my end. Once I got the server restarted (with some config changes), the DNS issue is solved, and I can reach Adobe again. They have changed their bugbase JSON responses slightly so I need to rejig my end of things, but that's nae bother and I should have it sorted out tomorrow.I really did leap to judge too soon here, and I feel a bit daft for having done so. I'm all for applying derision to a situation, but I do actually think it should have a foundation, and it doesn't in this case.
Apologies, Adobe.
I'm gonna leave the rest of the article here as a monument to my stupidity.
This time, let's nip something at the bud: I don't give a shit who interprets who is at fault for any of this, nor how people confuse the notions of fault and blame, and will contort analogies to convince themselves they're right (I am not blameless here). This isn't about that.
Adobe have messed with the viability of some community projects again. This time: my ones.
Yesterday morning I noticed the Adobe bugbase was down, and duly posted a nudge on Twitter to let them know. Some time later it was all up and running again. I didn't know how long it had been down - it'd've been at least 12h I think - so I decided to do a manual run of the processing for the @CfmlNotifier Twitter feed. This is two separate processes: one polls for any new bugs that have been raised since the last time the polling ran, and posts a Twitter update along these lines:
3737264 (CF 11) SecureProfile should not install Example Datasources, Gateways, or Solr Collections (http://t.co/vWPmYrtOpI)
— CfmlNotifier (@CfmlNotifier) April 5, 2014
I run this process every two hours. I also run another longer-winded process every six hours (I think?), which trawls through all open ColdFusion bugs, and see if there's been any activity on them: status changes; or new votes, comments or attachments, eg:
VOTE ADDED: 3737264 (11.0) SecureProfile should not install Example Datasources, Gateways, or Solr Collections (http://t.co/vWPmYrtOpI)
— CfmlNotifier (@CfmlNotifier) April 5, 2014
This has to hit the bug tracker a few hundred times as it checks each bug. There is no way that I have been able to guess to do one query to get all the relevant info: I need to query (via HTTP) each bug separately. This process takes about 15min to run.
Anyway... neither of those processes would run any more: each getting an HTTP timeout on the first request.
Initially I thought it was something to do with my account: it's a freebie / shared server account at CFMLDeveloper, so thought perhaps Russ had locked something down, or I had used too much bandwidth (unlikely: all this stuff is very low traffic). I tried CFHTTPing other domains, and those were all fine.
Next I tried making a <CFHTTP> request to the bugbase - one of the requests that was timing out when I ran the notification update process - from my own PC, and that ran fine. I also asked on the ColdFusion IRC channel if someone there - sharondio stepped up - and ran my code in their local environment, and the request went through fine for them too.
I shelved troubleshooting all this at that point as the day was dragging on and I had food to eat, and mindless TV to watch.
This morning I set up all the relevant code on my machine here, and ran it, and it all ran fine. Adobe have also tweaked the JSON coming back from one of the requests, but not in a significant way (not in a sensible way either, I hasten to add, but hey), and I altered my code accordingly.
So it looks to me like Adobe have blocked access to their bug base from my rig @ CFMLDeveloper. Which is interesting. What it means is the following things no longer work:
- Twitter status updates to @CfmlNotifier for new/updated bugs;
- the "50-most recent ColdFusion Bugs" RSS feed;
- its pal the "50-most recent ColdFusion Builder Bugs" RSS feed;
- the wee bug tracker search thing I knocked together;
- the "Untriaged ColdFusion 10 Bugs List".
The good news is... and sorry to bury this down the bottom... is that I actually started getting notifications of updates made to bugs I had participated in (voting, etc) from Adobe this morning. Here's an example:
Dear User,
Bug 3316802 for the product "ColdFusion" that you submitted has changed. You can click 3316802 to view or update the bug.
Thank you,
Adobe
This came through within a few minutes of me adding a comment to that ticket. This was a bug I myself had raised (well: it's not as bug, and Adobe should not refer to it as such: it was very much an enhancement request). But I also commented on a coupla bugs I hadn't myself raised, and I got notifications for those ones too. So this is great news.
Initially I thought this might remove the need for the @CfmlNotifier feed, but on reflection I think it covers different ground, so it's a shame it's not working any more. It's different because one will only get update emails for bugs one has already participated in, whereas @CfmlNotifier sends updates for all bugs (well: all new bugs and enhancement requests, and updates to open bugs / E/Rs). And that's still useful info that Adobe do not themselves furnish.
I will touch base with Adobe as to why they seem to be blocking me (and indeed if this is the case, but it certainly seems that way), and if they can reverse that. It might be their firewall deciding I'm hitting them too often, or something. Who knows?
Anyway, sorry that stuff is down. I will still run the @CfmlNotifier updates manually a coupla times a day - before and after work, or if I think about it on the weekend - but the RSS feeds and the search UI will continue to be dead for the foreseeable future. Such is life.
--
Adam